at SciVerse ScienceDirect

Journal of Archaeological Science 40 (2013) 1602e1610

Contents lists available

Journal of Archaeological Science

journal homepage: http: / /www.elsevier .com/locate/ jas

Provenance of marble sculptures from the National Museum of Carthage (Tunisia)

Nejia Laridhi Ouazaa a, Lluís Casas b,*, Aureli Álvarez b, Boutheina Fouzai a, Marta Moreno-Vide c,Laurence Vidal d, Roudosli Sihem e, Corinne Sonzogni d, Daniel Borschneck d

aUniversité de Tunis El Manar, Faculté des Sciences, Unité de Recherche, Pétrologie Cristalline et Sédimentaire, Département de Géologie, Campus Universitaire, 2092 Manar II,Tunis, TunisiabUniversitat Autònoma de Barcelona, Facultat de Ciències, Departament de Geologia, Campus de la UAB, 08193 Bellaterra, Catalonia, SpaincUniversitat Autònoma de Barcelona, Facultat de Filosofia i Lletres, Departament d’Art, Campus de la UAB, 08193 Bellaterra, Catalonia, SpaindCEREGE e CNRS UMR 7330, Aix-Marseille Université, Europôle Méditerranéen de l’Arbois, BP 80, 13545 Aix en Provence Cedex 4, Francee Institut National du Patrimoine 4, Place du Château, 1008 Tunis, Tunisia

a r t i c l e i n f o

Article history:Received 9 July 2012Received in revised form8 October 2012Accepted 22 October 2012

Keywords:ProvenanceArchaeometryStable isotopesWhite marblesCarthagePetrography

* Corresponding author. Tel.: þ34 935868365; fax:E-mail address: Lluis.Casas@uab.cat (L. Casas).

0305-4403/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.jas.2012.10.035

a b s t r a c t

The results of a combined minero-petrographic and isotopic study carried out on representative marblesculptures from the Museum of Carthage are presented. Pentelikon source has been identified for nine ofthe analyzed items, Carrara and Prokonnesos sources have been also identified for three and one itemsrespectively, confirming in some cases previously-made hypothesis. Local Chemtou and Djebel Ichkeulstones present very different petrographic fabrics compared with the most recurrent marbles. Theundertaken approach adds value to the descriptions of the studied artifacts and could be extended toother art pieces of this and other museums.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

The National Museum of Carthage is, along with the BardoMuseum, one of the main archaeological museums in Tunisia. Themuseum locates on the Carthage acropolis, on top of the Byrsa hill. Itwas founded at the endof the 19th century (Gandolphe,1952) by thePères Blancs (a society of Catholic missionaries), its building wasoriginally amonastery, and indeed, thefirst archaeologicalworks onthe site were performed by the missionaries. After the indepen-dence of Tunisia from France, and the decommissioning of thereligious functions of the building, themuseumopened as a nationalmuseum in 1964. It was refurbished at the end of the 20th centuryand it is at present undergoing again major restructuring.

The history of the Carthage Museum is only the last chapter ofthe full story of Byrsa hill and its surroundings. On top of this hill,known as l’esplanade, there was, once on a time, the Punic city ofCarthage (prosperous from the 6th century BC onwards) that wascompletely destroyed by the Romans after the Third Punic War(146 BC). A century later, a new Roman Carthage was built on itsruins under the rule of Julius Caesar. The Roman city became one of

þ34 935811263.

All rights reserved.

the biggest cities of the Empire and a center of early Christianity(4th to 7th century AD). Under the Vandals and the Muslims, thecity was progressively abandoned, used as a quarry and finallyforgotten until the arrival of the French missionaries.

Thought some of the finest finds from Carthage are now at theBardo Museum and elsewhere, the National Museum of Carthagecontains the largest collection of objects from this archaeologicalsite, covering both the Punic and the Roman periods. The Punicobjects include tomb and votive stones, inscriptions, sarcophagi,ossuaries and grave goods. From the Roman period there are mainlymosaics, funerary inscriptions, sculptures and architecturalelements. Finally, among the early Christian objects are worth tomention the funerary inscriptions and sarcophagi.

The scope of this paper is to bring archaeometric data to iden-tify, and in some cases to confirm, the provenance of several arti-facts from the Carthage Museum. The presented approach will helpto produce new explanatory notes for the visitors and could beextended to other art objects from the Museum.

2. Archaeological background

As a result of the old and complex history of the CarthageMuseum (Ennabli, 1998; Gandolphe, 1952), some of the objects

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e1610 1603

from its collections lack of a complete description of its archaeo-logical setting. Basic information, as the provenance of the stone-made pieces is not mentioned or assumed without any rigorousarchaeometric approach.

Themarbles of a number of sculptures, see Fig. 1, (including low-reliefs and sculpted sarcophagi) have been subjected to petro-graphic, X-ray and isotopic analyses for identification of itsprovenance.

All the studied statues and low-reliefs belonged to the Romancity. From the iconographic point of view, they reflect the abun-dance and nourishing vocation of the city: three sculptures includerepresentations of overflowing edibles as a symbol of wealth. Item#1 depicts a Genius Coloniae Cathaginis, i.e. a protecting spirit forthe Carthaginians, holding a cornucopia or horn of plenty. Thissculpture was found not far from the alleged emplacement of theCapitolium and presents all the features of the official imperial artfrom early 2nd century AD. Its face has some similarities with theknown representations of Antinous (the favorite of the Romanemperor Hadrian). Item #2 is a torso of a standing woman holdingwhat seems to be a basket full of fruits. The statue could representPomona, the goddess of fruit, usually associated to abundance. Item#3 is one of two colossal (3-m high) fine low reliefs from theMuseum depicting Victoria, the goddess of victory, as a wingedwoman holding a cornucopia.

The two similar low-reliefs of Victoria were found scattered inmany fragments by Delattre (1898) at the end of the 19th centurynearby the Saint Louis cathedral that still today stands on top theByrsa hill. Along with these low-reliefs, it was found item #4, alsoexceedingly fragmented, this one after reconstruction yieldedanother 3-m-high low relief of even finer manufacture. It depictsagain Victoria, now as a winged woman holding a trophy of Romanarms. This statue is considered one of the masterpieces of themuseum, and together with the other mentioned low-reliefs, hasbeen interpreted to be part of a triumphal arch or a similarmonument, perhaps to commemorate the victory over theParthians in 166 AD (Picard, 1951).

Other mythological characters from the studied set includeSilenius (item #5), the old fat satyr who tutored the wine-godBacchus. The statue is a free-standing sculpture that depictsa drunken Silenius being supported by four satyrs. The statue datesfrom the 1ste3rd century AD, perhaps is a reproduction of a biggerstatue and surely is an adaptation to a free-standing sculpture ofthis classical theme (Saumagne et al., 1964). It was found broken in45 fragments and after reconstruction two of the satyrs are stillmissing. According to the excavators, it appeared intentionallybroken (Saumagne et al., 1964), perhaps as a victim during theChristian destruction of pagan temples, in 399 (Lavagne, 2000).Item #6 is very incomplete, it was found in the 19th century behindthe amphitheatre (Delattre, 1899) and consists of a torso fragmentthat could depict a young Bacchus or perhaps Apollo. Finally, item#7 is a Roman copy, possibly from the 3rd century AD, of a Helle-nistic statue of a sleeping Eros (the god of love).

Besides mythological characters, other Roman sculptures fromthe studied set depict famous and anonymous people. Thus, item#8 portrays the head of co-emperor Lucius Verus, who ruled withMarcus Aurelius, from 161 AD to 169 AD, it was found during the19th century at the southern side of Byrsa hill (Delattre, 1899). Item#9 and #10 are funerary statues that were found within the samemausoleum, in the Yasmina necropolis, 200 m south-west from theemplacement of the Carthage circus. Item #9 was found in 1981(Annabi, 1981) and it is usually referred as an auriga (a slave thatdrove a biga chariot) though he could be an agitator (quadrigadriver) as well. However, taking into account, among other features,that he is holding a small amphora, an alternative and plausibleinterpretation has turned him into a sparsor (stable hand who

cooled Roman chariot horses with water) (Thuillier, 1999). In 1992,item #10 was retrieved and it is usually assumed that it depicts thespouse of the sparsor (Norman, 1993; Norman and Haeckl, 1993).Both sculptures would be from middle 3rd century AD. Item #11 isa statue of an idealized woman wearing a stola sculpted using theflowing drapery technique, and item #12 is the torso of anunknown figure wearing distinguished clothes.

In addition to Roman sculptures, materials from two sarcophagiwere also studied. On one hand, item #13, known as the Sarcoph-agus of the Priest, depicts, carved in Greek style (Picard, 1951) on itslid, a bearded figure wearing a beltless tunic and thick sandals. Thissarcophagus was retrieved in 1902 from the Punic necropolis inBordj-Djedid (near Sainte Monique) along with the so-calledSarcophagus of the Priestess (Delattre, 1903a); both are centralpieces of the Punic-dedicated area of the museum, other similarsarcophagi were found in the same necropolis, all of them aresupposedly from the late 4th century BC to early 3rd century BC.Another simpler sarcophagus, item #14, was also studied. This isdecorated with parallel s-shaped strigils like many othersarcophagi found in the early Christian basilicas of the site(Delattre, 1912; Béjaoui, 1985, 1987). These sarcophagi date fromlate-4th to early 5th century and have been described as beingmade either of local biosparitic limestone ‘kadel’ or Prokonnesosmarble (Rodà, 2001).

3. Materials and methods

Small samples, in form of irregular chips of about 1 cm3, wereobtained from the fourteen listed items using a small chisel, alwaysfrom hidden parts of the objects (the back side or the outer bottom)to minimize the damage on the artifacts. Besides samples from theartifacts, samples from two local quarries of recrystallized lime-stones (Chemtou, in northwestern Tunisia and Djebel Ichkeul, innorthern Tunisia) were also obtained for comparison purposes. Partof the samples was powdered to be analyzed using both powder X-ray diffraction (XRD) andmass spectrometry. The remaining part ofeach sample was used for visual inspection and observation undera stereoscopic microscope and to produce thin sections to beviewed using a petrographic microscope. All samples were labeledaccording to the corresponding item number; samples fromChemtou and Djebel Ichkeul were labeled Ch and Dj respectively.

3.1. Minero-petrographic analyses

Standard thin sections of the samples were prepared at CEREGE.Petrographic descriptions were mainly performed using a NIKONEclipse 50iPOL microscope in the Institut d’Arqueologia Clàssica, inTarragona, pictures were taken using a NIKON COOLPIX5400mounted on the microscope. Reference materials from the collec-tion of classical marbles assembled by the LEMLA (Álvarez andRodà, 1992) in the Universitat Autònoma in Barcelona were usedfor comparison purposes. Thin section observation permits todetermine the fabric of themarble and to identify the accessory andsecondary minerals, in addition to calcite. Analyses performed byexperienced petrologists are known to be a successful method toidentify different types of marbles and related rocks.

3.2. XRD analyses

XRD measurements were performed at CEREGE on powderedsamples with a q-q Panalytical X’Pert Pro MPD diffractometerrunning at 40 kV and 40 mA using Co Ka radiation (l¼ 1.79�A) witha linear X’Celerator detector and a secondary flat monochromator.The 2q range was 23e75� with a step size of 0.033� and a countingtime of 3 s per step. Samples were finely crushed in an agatemortar

Fig. 1. Pictures of all the studied artifacts from the Museum of Carthage, numbers correspond to labels as defined in Section 2 and Table 1.

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e16101604

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e1610 1605

and sieved on “zero-background” silicon plate to avoid preferentialorientation. Particles statistic was improved by spinning thesamples at 15 rpm. Rietveld refinements (Rietveld, 1969) wereperformed in order to obtain semi-quantitative information aboutthe mineral phases present in the samples. Rietveld analysisconsists in simulating a diffraction pattern from the crystallo-graphic parameters of each mineral phase (space group, unit cell),peak profile parameters and some instrumental parameters. Thetheoretical pattern is then iteratively compared to the experimentalone adjusting the parameters to minimize the differences betweenthe two patterns. X’Pert Highscore Plus (Panalytical) was used toobtain the contribution of each mineral phase fitting only thescale factor and the background. Maud software was also usedrefining furthermore some instrumental, profile and structuralparameters. In particular, our target was the identification ofdolomitic components.

3.3. Isotopic analyses

Isotopes analysis (d18O and d13C) were conducted at CEREGE oncrushedmarble samples using a mass spectrometer (Finnigan DeltaAdvantage) equipped with a carbonate device. Measured isotopicvalues are normalized against international standard NBS-19. Meanexternal reproducibility was better than 0.03& for d13C and 0.05&for d18O. Isotopic characterization has proven to be a very useful toolto identify marble provenance in artefacts (Gorgoni et al., 2002),specially when its use is combined with minero-petrographicresults (Gorgoni et al., 2002; Pensabene et al., 2012). The isotopicsignature is a measure of the deviation of 13C/12C and 18O/16O ratiosfound in the investigated sample with respect to a standard sampleVDPB (Belemnitella americana from the Cretaceous Pee DeeFormation, South Carolina) (Capedri et al., 2004). The principalfactors controlling the isotopic ratios in marbles are linked to the

Table 1Analytical features of the materials sampled (all from the National Museum of Carthage bMGS ¼ maximum grain size. % Dol ¼ Dolomite percentage. Attr. ¼ Attribution. Pe ¼ Pen

Object Item Petrography

Textural features

Genius Coloniae 1 HE, mosaic of lineated crystals, strained crystals,occasional microcrystal aggregates.

Pomona 2 HO, polygonal non-oriented, straight/curvedboundaries, triple points.

Victoria withcornucopia

3 HE, mosaic of slightly lineated crystals, elongatedmicrocrystal aggregates.

Victoria with arms 4 HE, mosaic of slightly lineated crystals, lamellarclusters with carbonate and micas.

Silenius 5 HO, polygonal non-oriented, straight/curvedboundaries, triple points.

Bacchus 6 HE, mosaic of slightly lineated crystals, curbedboundaries.

Eros 7 HE, dispersed and aggregated microcrystals,zoned rhomboid crystals (dolomite).

Lucius Verus 8 HE, mosaic of slightly lineated crystals,curbed boundaries.

Sparsor 9 HE, mosaic of slightly lineated crystals,microcrystal aggregates.

Sparsor’s wife 10 HE, mosaic of slightly lineated crystals.Woman with stola 11 HE, polygonal non-oriented, straight/curved

boundaries, triple points.Unknown torso 12 HE, mosaic of slightly lineated crystals,

occasional microcrystal aggregates.The Priest 13 HE, mosaic of slightly lineated crystals.Christian

sarcophagus14 HE, xenoblastic structures, sutured boundaries,

intense twinning, strained crystals and twinning linesChemtou sample Ch HO, microcrystalline, anhedral grains, twinning,

occasional fine-grained crystals in veins.Ichkeul sample Dj HE, very fine grained, embayed boundaries.

nature of the protolith, the temperature and interactionwith waterduring metamorphism and later weathering history.

4. Results

Table 1 lists the essential features of the analyzed marble arti-facts obtained from the different applied methods.

The minero-petrographic analyses of thin sections (Fig. 2)revealed textural and mineralogical information. Regarding thecomposition, calcite is plainly the main, if not the only, componentin all samples. Rounded quartz crystals have been identified insamples #1, #4, #6, #9, #10, #12 and #13, whereas elongated micacrystals (possible muscovite) have been found in samples #4, #6and #12. Finally the presence of dolomite was also distinctive insample #7 (in form of zoned rhomboid crystals, Fig. 3a) andperhaps in sample #4 (in form of lamellar crystals with micas,Fig. 3b). As far as texture is concerned, all the samples, exceptone (sample #14), present mosaics of fine-grained crystals(MGS< 1mm). A systematic counting of several digitized images ofthin sections was performed to obtain crystal size distributions(CSD). Almost all samples exhibit rather large heteroblastic distri-butions covering at least the 0.15 mme0.60 mm range, generallywith slightly lineated crystals and sometimes (samples #1, #3, #7,#9 and #12) containing microcrystal aggregates (Fig. 3c) ofmicrospar size (up to 20 mm). In contrast, samples #2 and #5 showno microcrystals at all and a narrower CSD (w0.12 mme

w0.42 mm) of equidimensional crystals with occasional triplepoint boundaries (Fig. 3d), which can be called as homeoblastic.Finally, sample #14, definitively different from the rest, has a widerCSD, from w0.10 mm to w2.28 mm. Other identified distinctivefeatures are occasional strained crystals with undulating extinc-tions (samples #1 and #14) and intense twinning (again sample#14). Fabrics of samples from local quarries exhibit microcrystalline

ut items Ch and Dj). XRD ¼ X-ray diffraction. HE ¼ heteroblastic. HO ¼ homoblastic.telikon. C ¼ Carrara. Pr ¼ Prokonnesos.

XRD Isotopic ratio Attr.

Accessory minerals MGS (mm) % Dol d13C d18O

Quartz 0.67 Traces 2.63 �7.29 Pe

e 0.42 e 2.21 �1.71 C

e 0.60 6 2.64 �4.37 Pe

Muscovite, quartz 0.55 Traces 3.25 �5.55 Pe

e 0.43 Traces 2.12 �2.21 C

Quartz, mica 0.60 Traces 2.43 �7.56 Pe

e 0.60 12 2.85 �6.48 Pe

e 0.77 1 2.57 �4.17 Pe

Quartz 0.61 2 2.36 �6.17 Pe

Quartz 0.72 e 2.58 �4.96 Pee 0.62 e 2.14 �1.67 C

Quartz, muscovite 0.61 12 2.97 �5.55 Pe

Quartz 0.85 2 2.80 �7.77 Pe

.e 2.28 e 3.28 �0.65 Pr

e 0.04 e 1.24 �10.45 e

Opaque minerals 0.38 Traces 1.46 �6.46 e

Fig. 2. Crossed polarized light photomicrographs of the thin sections of some of the objects studied. Numbers correspond to labels as defined in Section 2 and Table 1.

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e16101606

equidimensional anhedral calcite crystals with veins filled withcoarser grains (Fig. 4a) and a very fine-grainedmosaic of calcite andopaque crystals with embayed boundaries (Fig. 4b).

XRD analyses, in agreement with petrographic analyses, revealthat calcite is the main component of all the samples (representa-tive patterns are shown in Fig. 5). However, there are slightdiscrepancies between XRD and petrographic data regarding theaccessory minerals: XRD diffractograms indicate an important

Fig. 3. Crossed polarized light photomicrographs of detailed fabric features found in the thdolomite; (b) sample #4 e lamellar crystals of carbonate; (c) sample #9 e microcrystal ag

fraction of dolomite in samples #7 (Fig. 5b) and #12 (12%) andsample #3 (6%), though petrographic analyses only detected it insample #7 (Fig. 3a) and thus, they went unnoticed in samples #3and #12. In contrast, presumed dolomite lamellar crystals insample #4 could be, after all, calcite crystals. XRD patterns haverevealed the presence of quartz in all the samples, but only ina significant amount in samples #9 and #13 (2%) and #6 and #12(1%), though quartz crystals were also seen in the thin sections of

in sections of some of the objects studied: (a) sample #7 e zoned rhomboid crystal ofgregate; (d) sample #2 e homeoblastic mosaic with triple points.

Fig. 4. Crossed polarized light photomicrographs of the thin sections of samples from two Tunisian quarries: (a) Chemtou and (b) Djebel Ichkeul.

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e1610 1607

samples #1, #4 and #10 and finally muscovite traces have beenfound in samples #4, #6 and #9, though petrographically thismineral went unnoticed in sample #9 but it appeared in thinsections of sample #12. Discrepancies between XRD and petro-graphic analyses reflect the natural variability of accessorycomponents in the studied samples. In fact, the powdered XRDsamples were not obtained from thematerial immediately adjacentto the cut slab.

The isotope ratios (d18O and d13C) of the samples are shown inthe diagrams of Fig. 6. According to the petrographic data, all thesamples except one (#14) are plotted in the comparative diagramfor fine-grained classical marble varieties (Gorgoni et al., 2002),

Fig. 5. Indexed X-ray diffraction patterns for (a) sample #10 e all the peaks correspondto calcite except the quartz-labeled (101) which is the main reflection of quartz; (b)sample #7 e same peaks as in sample #10 and extra dolomite-labeled peaks (in italics)that correspond to dolomite.

whereas sample #14 is plotted separately in the diagram formedium- to coarse-grained classical varieties (Gorgoni et al., 2002).The majority of samples lie unambiguously within a given area,corresponding either to Carrara marbles (for samples #2, #5 and#11) or to Pentelikon marbles (for the rest of the samples, thoughsample #8 could also correspond to a Dokimeian marble). Finally,the coarse-grained sample (#14) should belong to a Prokonnesos-1marble (from the island of Marmara) or a Thasos-2 marble (fromthe Aliki district, in the island of Thasos). Moreover, regardless oftheir grain size, all the samples have been also plotted ina comparative diagram with the isotopic areas of several knownwhite marbles from Roman Hispania (Lapuente et al., 2000) andMorocco (Origlia et al., 2012; Antonelli et al., 2009a). All samples,except #6 and #13 lie within a given area of Hispanic marbles,sometimes with ambiguity between two different provenances.Samples #2, #5, #11 and #14 lie within the area of the Málagamarble district, samples #3, #8 and #10 within the area of theMacael group, in the Almeria marble district, though sample #10,along with samples #1, #4, #7, #9 and #12, also lie within the areaof the Estremoz marble district. Finally samples #7 and #9 are alsowithin the area of the Almadén de la Plata marbles.

5. Discussion

The use of isotope diagrams is a good starting point to constraintthe possible origin of the studied marbles. Considering classicalmarbles, our isotopic results indicate quite unambiguously thatnine of fourteen investigated items are made of Pentelikon marble,three of them are made of Carrara marble (Fig. 6a), and only oneitem (#14) is ambiguously identified as a Prokonnesos-1 or Thasos-2 marble (Fig. 6b). Besides classical marbles, we have consideredRoman Hispania marbles because some of its marbles have beenidentified in African sites (Antonelli et al., 2009a; Origlia et al.,2011) and several researchers have hypothesised that Hispanicmarbles were not only used in Iberia but also exported (Morbidelliet al., 2007). We have also considered Moroccan marbles forproximity reasons though its isotopic signature resulted to bedifferent from that of our studied marbles. Despite being quarriedeven closer to Carthage, we have disregarded the greco scrittomarble (from Cap de Garde, Algeria) because its fine gray-blackishveining and dark gray-to-bluish patches/spots (Antonelli et al.,2009b) are absent in all our studied items.

The overlapping between the isotopic signature of classical andHispanic marbles is a problem that has to be solved taking intoaccount data from other techniques, mainly petrography. Indeed,the Hispanic origin of the samples can be excluded due to petro-graphic and mineralogical reasons. For the samples that lie withinthe area of the Málaga marble district (#2, #5, #11 and #14), thisorigin can be rejected because Málaga marbles are invariablydolomitic (Lapuente et al., 2000) and our samples are calcitic, thus

Fig. 6. Isotopic signatures of white marbles used for the studied artifacts; isotopicfields from Gorgoni et al. (2002), Lapuente et al. (2000) and Origlia et al. (2012). (a) Pa-1 ¼ Paros Stefani; D ¼ Dokimeion; Pe ¼ Pentelikon; C ¼ Carrara; (b)Aph ¼ Aphrodisias; N ¼ Naxos; Pa ¼ Paros; Pr ¼ Prokonnesos; T ¼ Thasos; (c) HispanicAE ¼ Estremoz Anticline district. Ossa Morena Zone district: VA ¼ Viana do Alentejo,Al ¼ Alconera, AM ¼ Almadén de la Plata. Almeria district: Lu ¼ Lubrin, Mc ¼ Macaelgroup. Ma ¼ Malaga district. AG ¼ Alhaurin el Grande; and Moroccan T ¼ Tiskram;SL ¼ Sidi Lamine; HI ¼ Hrent Ifou and BA ¼ Bou Acila.

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e16101608

these samples would have a Carrara provenance (except for thecoarse-grained #14). Moreover, samples within the isotopic area ofthe Estremoz district (#1, #4, #7, #9, #10 and #12) are fine-grained,whereas almost all lithotypes from the Estremoz district aremedium- or coarse-grained, and often with pinkish veins (Álvarezet al., 2009). Among the Estremoz varieties, only the marblesfrom Borba are fine-grained, but they contain gray veins (Álvarezet al., 2009) and contrary to our samples, they have not lineated

crystals and are texturally closer to Carrara marbles (Lapuente,1997). Thus, Mount Pentelikon remains as the most probableprovenance for these samples. Hypothetical Macael origin forsamples #3, #8 and #9 can also be excluded because the largergrain size of the Macael marbles compared with our samples(Lapuente et al., 2000; Álvarez et al., 2009). Finally, marbles fromAlmadén de la Plata have a distinctive mortar and strained texture(Lapuente et al., 2000), very dissimilar to the samples that matchtheir isotopic signature (#7 and #9).

Once excluded the Hispanic marbles, we can keep the resultsobtained from the isotopic diagrams of classical marbles, these ingeneral point to Pentelikon and Carrara. The petrographic featuresof our samples are in agreement with these isotopic attributions.Despite being marbles with a similar tectonic setting (regionalmetamorphism), Pentelikon and Carrara have distinct isotopicsignatures, for instance isotopic values for Carrara marbles aremuchmore clustered indicating a static recrystallization that is alsoapparent by a petrographic homeoblastic texture.

The occurrence of Hispanic marbles in Carthage would havebeen surprising but the occurrence of both Pentelikon and Carraramarbles in Carthage during Roman times was expected. White-marble formations are quite scattered in the Mediterranean basinand some of them have been quarried since Antiquity, among themost renowned we find certainly those from Pentelikon and Car-rara. The quarries of Mount Pentelikon (Greece) were worked fromthe 5th century BC (Korres, 1995) and from that time onwards itsmarbles were freely exported. Pentelikon exports reached Rome asearly as the 2nd century BC and though its quarries never becamean imperial property (Attanasio et al., 2006), it was widely used anddistributed across the Roman Empire (Herrmann et al., 2009),including distant areas such as north-west Africa (Antonelli et al.,2009a). In Carthage, for instance, it was applied in the bases andthe capitals of the Antonin thermae (2nd century AD) (Duval, 1971).Concerning Carrara, Pliny the Elder explained, in his Natural History(André et al., 1981), that the first works in the Carrara quarry weredone in Julius Caesar’s times (1st century BC) and soon became oneof the main marble sources both for sculpture and buildingpurposes and its exports reached the Roman colonies in Africa(Azedine, 2004) as well. We cannot make any chronologicaldistinction between the use of Pentelikon or Carrara marbles. Bothtypes of marbles were used in Rome preferentially during the 1stand 2nd centuries AD (Bruno et al., 2002) and quite consistently,the ages of the items according to artistic/typological featureswould be 2nd and 3rd centuries AD.

As previously mentioned, the overlapping between the fields ofvariability of the two isotopic parameters (d18O and d13C) has beenone of the problems of isotopic analysis applied to provenancestudies. For classical marbles, this problem has been overcome bytaking account of data from other techniques, such as themaximum grain size (MGS). However some overlapping persists. Inthe present study, sample #8, and perhaps #3 and #10 too, could beidentified as Dokimeian as well as Pentelikon marble. Again, thepetrographic information is essential to make the distinction,Dokimeion marble is fine-grained as Pentelikon but its fabric iscompletely different: the crystal boundaries are often sutured andinclude strained, often kinked, crystals indicating unstable condi-tions reached after quite brief metamorphic events (Gorgoni et al.,2002). Thus, samples #8, #3 and #10 are Pentelikon marbles andnot Dokimeian.

Sample #13 deserves a separate mention because its origin isnot Roman but Punic. According to the isotopic analysis this sampleis made of Pentelikon marble. However, due to its pale yellow tintand its Punic origin we wondered if this could be made of a localstone. Our candidates were the yellowish recrystallized limestonesfrom Chemtou and Djebel Ichkeul quarries. Looking at the isotopic

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e1610 1609

ratios obtained from samples from these two quarries (see Table 1)we get values that lie within the field of variability of Pentelikonmarble, though with quite low values of d13C. Other data sourcescan help to identify the provenance of sample #13 marble. On onehand, Chemtou can be excluded because of historical reasons: itsquarry produced the renowned giallo antico (Roman marmornumidicum) but it was worked from the second half of the 2ndcentury BC (Mezzolani, 2008) and thus, long after the presumedage of item #13 (the Punic sarcophagus), moreover its fabric isdefinitively not similar to sample #13. On the other hand DjebelIchkeul, although it was quarried during Punic times (Mezzolani,2008), can also be excluded taking into account petrographicdata: the fabric of sample #13 is unequivocally that of a fine-grained marble (a material that has undergo complete recrystalli-zation due to metamorphism) whereas the stone from Djebel Ich-keul, like that from Chemtou, is an intermediate type of rockbetween the sedimentary and metamorphic types. It has exten-sively recrystallized through diagenesis and low-grade meta-morphism. Recrystallization has resulted in a marble-like fabric butwith a significant lower grain size (see Table 1) compared to fine-grained marbles. The question remains as to where the yellowishtint of item #13 comes from. This could, in fact, be compatible withthe Pentelikon attribution for some authors have referred on Pen-telikon marble a golden tint due to weathering of iron compounds(Gardner, 1920). However, the opaque components seen in the thinsections of sample #13 are very small (<20 mm) and unseen in thediffraction patterns, and thus possibly amorphous. An extra sourceof tint could be the wooden coffins that lay on the sarcophagus as ithas been pointed out for other items from the same site (Delattre,1903b). In conclusion, sample #13 could be made of Pentelikonmarble, its Punic origin is not an objection to this attribution. ThePunic culture was strongly influenced by the Greeks; this has beenspecially remarked after revision of the findings from the necrop-olis in Bordj-Djedid where item #13 was found. Sarcophagi fromthis necropolis are carved in Greek style and its possible Greekorigin was already hypothesized by Picard (1951).

Another marble with a different background is that of item #14.Petrographically, it is a marble with grains coarser than those of allother items and thus, its isotopic results are plotted in a differentcomparative diagram for classical marbles. Isotopic analyses pointto two possible origins (Prokonnesos-1 or Thasos-2). Once more,petrographic data are the clue to narrow the attribution. The Tha-sian origin can be excluded because MGS for sample #14 is below2.5 mm (Palagia et al., 2003). Moreover sample #14 shows xeno-blastic structure with some crystals granulated along their borders(mortar texture), sutured boundaries and deformation elementsthat match with the typical features of Prokonnesos-1 marbles(Gorgoni et al., 2002). However, the lack of the typical gray veins ofProkonnesian could cast some doubts on this attribution, thoughthese characteristic stripes are sometimes absent in Prokonnesos(Attanasio et al., 2008). Anyhow, this attribution is consistent withthe early Christian character of the item as Prokonnesos largelyreplaced other marbles from the 2nd century BC onwards(Attanasio et al., 2006, 2008).

6. Conclusions

An archaeometric approach has been undertaken to determinescientifically the provenance of some artifacts from the Museum ofCarthage. The obtained data add value to the characterization anddescriptions of the studied items and the undertaken approachcould be extended to other art pieces. Analytical results indicatethat the most probable source for nine of the fourteen investigateditems is Pentelikon marble, for three of them is Carrara and onlyone would be Prokonnesos-1 marble. These sources are consistent

with the historical context and facts, and some authors had alreadycorrectly presumed them for some items: e.g. item #5 (Saumagneet al., 1964), item #9 (Thuillier, 1999) or item #14 (Rodà, 2001),even without a rigorous archaeometric approach.

Thanks to the compilation of isotopic databases, the fields ofisotopic fractionation variability (d18O and d13C) for the mostrecurrent marbles seem to be now well-established. This allowsa fast identification of marble provenance that requires a very smallamount of sample. However, isotopic values alone are of poorsignificance as they do not even discriminate between pseudome-tamorphic carbonates (like Ch or Dj samples) and marbles. It isknown that metamorphism on carbonates can actually keep unal-tered, to a certain extent, their original sedimentary isotopicsignature (Valley et al., 1990). The combination of minero-petrographic and isotopic analyses has proved once again to bea reliable approach to identify the provenance of marbles.

Acknowledgments

We are grateful to Montserrat Claveria for helping in thedescription of the artefacts. Pierre Rochette is also recognized forhelping in the isotopic analyses. Financial support from the Unitéde recherche: Pétrologie cristalline et sédimentaire (05/UR/10-03),the Tunisian French Joint Committee for University Cooperation(Program PHC-Utique,11G/1004), the SpanishMinisterio de Cienciay Innovación (project HAR2010-16953), and the Agencia Españolade Cooperación Internacional para el Desarrollo (Spain-Tunisiabilateral project A1/039844/11) is acknowledged.

References

Álvarez, A., Domènech, A., Lapuente, P., Pitarch, A., Royo, H., 2009. Marbles andStones of Hispania: Catálogo Exposición, IX ASMOSIA Conference, Tarragona.Institut Català d’Arqueologia Clàssica.

Álvarez, A., Rodà, I., 1992. Ciencias, metodologías y técnicas aplicadas a la arqueo-logía. Fundació “la Caixa”.

André, J., Bloch, R., Rouveret, A., 1981. Pline l’ancien: histoire naturelle, Livre XXXVI.Les Belles Lettres.

Annabi, M.K., 1981. Découverte d’un édifice et d’une sculpture près du cirqueRomain. CEDAC Carthage Bulletin 3.

Antonelli, F., Lazzarini, L., Cancelliere, S., Dessandier, D., 2009a. Volubilis (Meknes,Morocco): archaeometric study of the white and coloured marbles imported inthe Roman age. Journal of Cultural Heritage 10, 116e123.

Antonelli, F., Lazzarini, L., Cancelliere, S., Dessandier, D., 2009b. Minero-petro-graphic and geochemical characterization of ‘Greco scritto’ marble from Cap deGarde, near Hippo Regius (Annaba, Algeria). Archaeometry 51, 351e365. http://dx.doi.org/10.1111/j.1475-4754.2008.00408.x.

Attanasio, D., Brilli, M., Bruno, M., 2008. The properties and identification of marblefrom Proconnesos (Marmara island, Turkey): a new data including isotopic, EPRand petrographic data. Archaeometry 50, 747e774.

Attanasio, D., Brilli, M., Ogle, N., 2006. The Isotopic Signature of Classical Marbles.“L’Erma” di Bretschneider.

Azedine, B., 2004. Aspects de l’hellénisme africo-romain. Comptes-rendus desséances de l’Académie des Inscriptions et Belles-Lettres 148, 53e65.

Béjaoui, F., 1987. Nouveaux sarcophages à strigiles de Carthage. CEDAC CarthageBulletin, 5e6.

Béjaoui, F., 1985. Sarcophages inédits de Tunisie. CEDAC Carthage Bulletin, 25e28.Bruno, M., Cancelliere, S., Gorgoni, C., Lazzarini, L., Pallante, P., Pensabene, P., 2002.

Provenance and distribution of white marbles in temples and public buildingsof Imperial Rome. In: Hermmann, J.J., Herz, N., Newman, R. (Eds.), Asmosia 5.Interdisciplinary Studies on Ancient Stone. Proceedings of the Fifth Interna-tional Conference of the Association for the Study of Marble and Other Stones inAntiquity. Boston, pp. 289e300.

Capedri, S., Venturelli, G., Photiades, A., 2004. Accessory minerals and d18O and d13Cof marbles from the Mediterranean area. Journal of Cultural Heritage 5, 27e47.

Delattre, A.L., Commission de l’Afrique du Nord (France), 1899. Musées et collectionsarchéologiques de l’Algérie et de la Tunisie: Musée Lavigerie de Saint Louis deCarthage. E. Leroux.

Delattre, A.L., 1912. Les fouilles de Damous-el-Karita. Comptes-rendus des séancesde l’Académie des Inscriptions et Belles-Lettres 56, 460e476.

Delattre, A.L., 1903a. Carthage. Nécropole punique voisine de Sainte-Monique. Deuxsarcophages anthropoïdes en marbre blanc. Comptes-rendus des séances del’Académie des Inscriptions et Belles-Lettres 47, 23e33.

Delattre, A.L., 1903b. Carthage. Nécropole punique voisine de Sainte-Monique. Leseptième et le huitième sarcophage de marbre. Couvercle anthropoïde. e

N.L. Ouazaa et al. / Journal of Archaeological Science 40 (2013) 1602e16101610

Épitaphes de prêtresses. Comptes-rendus des séances de l’Académie desInscriptions et Belles-Lettres 47, 11e23.

Delattre, A.L., 1898. frLes Grandes statues du Musée de Saint-Louis à Carthage. imp.E. Petithenry, Paris.

Duval, N., 1971. Architecture et urbanisme à Carthage et à Utique. Journal dessavants 4, 287e297.

Ennabli, A., 1998. Le musée de Carthage: un lieu de mémoire. Museum Interna-tional. n 198, L, 25e32.

Gandolphe, P.,1952. Origines et débuts duMusée Lavigerie. Cahiers de Byrsa 2,151e178.Gardner, E.A., 1920. A Handbook of Greek Sculpture. Macmillan, London.Gorgoni, C., Lazzarini, L., Pallante, P., Turi, B., 2002. An Updated and Detailed Min-

eropetrographic and CeO Stable Isotopic Reference Database for the MainMediterranean Marbles Used in Antiquity, pp. 115e131.

Herrmann Jr., J.J., Tykot, R.H., DePaepe, P., Moens, L., 2009. Pentelic marble in centralItaly: isotopic testing of neo-attic furniture. In: Jockey, P. (Ed.), Marbres et autresroches de la Méditerranée antique: études interdisciplinaires/InterdisciplinaryStudies on Mediterranean Ancient Marble and Stones. Maisonneuve & Larose,Paris, pp. 309e328.

Korres, M., 1995. From Pentelicon to the Parthenon: the Ancient Quarries and theStory of a Half-worked Column Capital of the First Marble Parthenon. Melissa.

Lapuente, M.P., 1997. Problemas petrográficos en la identificación de mármolesclásicos: diferenciación de Carrara y Borba. Caesaraugusta 73, 279e288.

Lapuente, M.P., Turi, B., Blanc, P., 2000. Marbles from Roman Hispania: stableisotope and cathodoluminiscence characterization. Applied Geochemistry 15,1469e1493.

Lavagne, H., 2000. La tolérance de l’Église et de l’État à l’égard des oeuvres d’art dupaganisme dans l’Antiquité tardive. Études littéraires 32, 45e54.

Mezzolani, A., 2008. I materiali lapidei nelle costruzioni di età fenicia e punicaa Cartagine. Analecta Romana Instituti Danici, XXXIII, pp. 7e25.

Morbidelli, P., Tucci, P., Imperatori, C., Polvorinos, A., Preite Martínez, M., Azzaro, E.,Hernández, M.J., 2007. Roman quarries of the Iberian peninsula: “Anasol” and“Anasol”-type. European Journal of Mineralogy 19, 125e135.

Norman, N.J., 1993. The University of Georgia in the Yasmina Necropolis. CEDACCarthage Bulletin 9.

Norman, N.J., Haeckl, A.E., 1993. The Yasmina necropolis at Carthage. Journal ofRoman Archaeology, 238e250.

Origlia, F., Gliozzo, E., Gandin, A., Meccheri, M., Spangenberg, J.E., TurbantiMemmi, I., 2012. Marbles and carbonate rocks from central Morocco: a petro-graphic, mineralogical and geochemical study. Environmental Earth Sciences66, 209e222.

Origlia, F., Gliozzo, E., Meccheri, M., Spangenberg, J.E., Turbanti Memmi, I., Papi, E.,2011. Mineralogical, petrographic and geochemical characterization of whiteand coloured Iberian marbles in the context of the provenancing of someartefacts from Thamusida (Kenitra, Morocco). European Journal of Mineralogy23, 857e869.

Palagia, O., Maniatis, Y., Dotsika, E., Kavoussanaki, D., 2003. New investigations onthe pedimental sculptures of the “Hieron” of Samothrace: a preliminary report.In: Proceedings of the 7th International Conference of Association for the Studyof Marble and Other Stones in Antiquity. BCH Suppl. 51. École françaised’Athènes, Athens, pp. 113e132.

Pensabene, P., Antonelli, F., Lazzarini, L., Cancelliere, S., 2012. Provenance of marblesculptures and artifacts from the so-called Canopus and other buildings of “VillaAdriana” (Hadrian’s villa e Tivoli, Italy). Journal of Archaeological Science 39,1331e1337.

Picard, C., 1951. Carthage, Société d’Édition. Les Belles Lettres, Paris.Rietveld, H.M., 1969. A profile refinement method for nuclear and magnetic

structures. Journal of Applied Crystallography 2, 65e71.Rodà, I., 2001. Producción, materiales y circulación de sarcófagos en el Imperio

Romano. In: Conde Guerri, M.E., Noguera Celdrán, J.M. (Eds.), El sarcófagoromano: contribuciones al estudio de su tipología, iconografía y centros deproducción. Universidad de Murcia, Murcia, pp. 51e78.

Saumagne, C., Poinssot, L., Pinard, M., 1964. Mélanges de Carthage: Section cultur-elle de l’Ambassade de France.

Thuillier, J.P., 1999. Agitator ou Sparsor? À propos d’une célèbre statue de Carthage.Comptes-rendus des séances de l’Académie des Inscriptions et Belles-Lettres143, 1081e1106.

Valley, J.W., Bohlen, S.R., Essene, E.J., Lamb, W., 1990. Metamorphism in the Adir-ondacks: II. The role of fluids. Journal of Petrology 31, 555e596.